Chatlog Keyword Analyzer

# ### Chatlog Keyword Analyzer by Append Huang
# Timestamp: 20220831 01:50 UTC+8
#
# Usage: python keyword_analyzer.py chat.log
# Use Chatty log (or, part of Chatty log) / Rechat log (txt format) as input
# Please save the log as chat.log or put the filename as a command-line arguments
#
# for detailed output, use:
# python keyword_analyzer.py chat.log 1

import sys
from collections import Counter

filename = "chat.log"
if len(sys.argv) > 1:
    filename = sys.argv[1]

VERBOSE = False
if len(sys.argv) > 2:
    VERBOSE = bool(int(sys.argv[2]))

with open(filename,'r', encoding='utf-8') as f:
    f_lst = f.readlines()

user_c = Counter()
msg_c = Counter()
user_msg_lst = []
for line in f_lst:
    line2 = line.strip().split(']',1)[1]
    if line2.startswith(" MOD_ACTION:"):
        # MOD_ACTION:  (automod_rejected [racism1; "Ching Chong":identity7] <luffymtlqpb> Ching Chong)
        try:
            line2 = line2.split(']',1)[1][:-1]

        # [23:29:53] MOD_ACTION: id9000 (emoteonly) --> skip this line
        except IndexError:
            continue

    def user_helper(x):
        return x.isalnum() or x=='(' or x==')' or x==' '

    user2, msg2 = "", ""
    if "<" in line2 and ">" in line2: # <nastasyaebu> colored
        user, msg = line2.split(">")
        user2 = user.split("<",1)[1].strip()
        user2 = "".join([x.lower() for x in user2 if user_helper(x)]) # remove symbols
        msg2 = msg.strip()

    elif ":" in line2: #[10:45:51.135] recilurn: 老二 老二 老二
        user, msg = line2.split(":", 1)
        user2 = user.strip()
        msg2 = msg.strip()

    if user2 and msg2:
        user_c[user2] += 1
        msg_c[msg2] +=1
        user_msg_lst.append([user2, msg2])

if VERBOSE:
    print("# User List:")
    for user_count in sorted(user_c.items(), key=lambda x:x[1], reverse=True):
        print("    %s: %d"%user_count)
    print()
    print("# Msg List:")
    for msg_count in sorted(msg_c.items(), key=lambda x:x[1], reverse=True):
        print("    %s: %d"%msg_count)
    print()

key_lst = [x for x,c in sorted(msg_c.items(), key=lambda x:x[1], reverse=True) if c >1]
user_key_lst = [(i,v) for i,v in user_msg_lst if v in key_lst]
user_c2 = Counter()
msg_c2 = Counter()
for user, msg in user_key_lst:
    user_c2[user] += 1
    msg_c2[msg] += 1

print("# User List with Repeated Keywords:")
for user_count in sorted(user_c2.items(), key=lambda x:x[1], reverse=True):
    print("    %s: %d"%user_count)
print()
print("# Msg List with Repeated Keywords:")
for msg_count in sorted(msg_c2.items(), key=lambda x:x[1], reverse=True):
    print("    %s: %d"%msg_count)